In the 2018 proposed Medicare Access and CHIP Reauthorization Act’s (MACRA’s) rule, published earlier this year, HHS has again proposed to exclude two-thirds of physicians, or 900,000, from participation in the Merit-based Incentive Payment System (MIPS). MIPS was created under 2015 MACRA legislation to incent financially Medicare physicians and other Medicare Part B clinicians to improve care quality and reduce Medicare spending growth. HHS is choosing to exclude smaller-sized physician practices because, it is believed, MIPS reporting requirements place too high a burden on less resourced practices. However, MACRA legislation includes a “virtual group” provision that allows for solo and small group practices to partner in meeting MIPS reporting requirements. For technical reasons the Centers for Medicare and Medicaid Services (CMS) was delayed in implementing the provision. The first year in which providers can participate in MIPS as a virtual group will be 2018. Designed correctly, virtual groups offer substantial advantages that include greater MIPS participation, more competitiveness, higher financial reward and more opportunity for small practices to keep their independence.
MACRA legislation allows for three ways for physicians or Eligible Clinicians (ECs) to participate in MIPS. These are as individual or solo ECs, as a group of ECs, or as “virtual groups” consisting of any number of solo ECs and/or groups of 10 or fewer ECs. As CMS states in its proposed 2018 MACRA rule, published this past June 30, virtual groups are, again, intended to provide individual and small groups “support and reduce [MIPS reporting] burden.” The 2018 proposed defines at length the definition of a virtual group, how virtual groups are to structured, how and when they can apply for virtual group status and other relevant information. Performance year 2018 will, again, constitute the first year of MIPS virtual group performance because in last year’s final MACRA rule, CMS declined to implement MACRA’s virtual group provision citing “significant barriers” in developing the supportive technological infrastructure.
In more detail, CMS is proposing to define a virtual group as solo or small groups if they do not fall below either of two MIPS low volume exclusion thresholds: $90,000 or less in Medicare charges; or, 200 or fewer Medicare beneficiaries served. (The proposed 2018 MIPS exclusion thresholds were discussed and evaluated in an August 3rd Health Affairs blog post by Introcaso.) Individuals or groups must select to participate in a virtual group before the start of the performance year and cannot change their election during that time. Virtual groups, CMS also proposes, will consist of a combination of an unlimited number of Tax Identification Numbers (TINS) with no geographic or clinical specialty restrictions. They must operate under a written agreement that CMS plans outline or define via proposed model language. Groups need to be formed by December 1 of the preceding year, for example, by December 1, 2017 for the 2018 performance year.
What is noticeably missing in the proposed rule is any description of how solo or small group practices are to identify desirable virtual group partners. The agency provides no direction or instruction in answering this essential and practical question. This is black box. Solo and small group practices are expected to somehow credibly identify any number of other eligible and similarly sized provider groups located anywhere in the country and partner with them such that they collectively maximize their MIPS score. Absent this ability, their alternative is to blindly form a virtual group in the hope that the effort involved in forming the group is offset by greater collective success. Either expectation is naïve. Evidence of this is found in the proposed rule. Based on stakeholder feedback, CMS estimates only 16 virtual groups will be formed for performance year 2018.
We believe therefore a proactive matchmaking process that facilitates the creation of virtual groups is required for the MIPS provision to work. Recognizing the importance of how virtual groups are created and sustained, this past May we introduced to CMS staff our preliminary thoughts concerning how the agency could use historical claims, quality performance metrics and other data to inform and motivate solo and small group ECs to form or join a virtual group. Here we again outline our proposal in the hope Medicare stakeholders will offer their input or help further define the idea.
An Outline of Our Proposal
The matchmaking solution we propose is intended to answer the initial question confronting each solo and small group practice. Can a virtual group be formed that would maximize the participating solo or small group’s MIPS score. Since virtual groups do not currently exist the first practical step is to identify solo and small practices eligible to participate in a group, or those that exceed either exclusion threshold. The next step, obviously far more difficult, is in identifying solo and small practices that collectively have a reasonable, or statistically probable, chance of attaining a MIPS score that again would be higher than they could attain independently.
Conceptually, the solution we are proposing is a neural network-based learning algorithm that combines or exploits multiple data sources to create score maximizing virtual groups. Below is a high-level illustration of a neural network based learning algorithm, or the major subsystems in our proposed solution. The input would include relevant claims data, historical Physician Quality Reporting System (PQRS), Value Modifier (VM) and Meaningful Use (MU) program data, related quality data and data from other sources available to CMS. The data would then be extracted and analyzed using tools already developed. The claims to measure predictor engine will take this extracted data and using neural network based algorithms predict quality and other measures. This engine will be constantly updated and improved. The resulting data would be accessed by micro services to create scorecards for CMS, or CMS vendors, to share with practices. The data could also be used to enhance CMS’ Quality Performance Payment (QPP) website to provide increasingly more targeted information to solo and small groups and/or can be exported to other systems like Tableau.
What the algorithm would be attempting to accomplish is to enable CMS to identify solo and small group providers that would be collectively advantaged by forming a virtual group. In statistical terms this essentially is the challenge of creating a hybrid regression analysis model. CMS, or a CMS vendor, would then contact the identified solo and small group providers that the algorithm demonstrates would be advantaged. Those contacted would then be free to choose to participate in or to form a virtual group. For those that choose to do so, CMS, or again a CMS vendor, would provide technical assistance or education and support. The immediate advantage a matchmaking model presents is it helps allow solo and group practices to avoid having to wait an extended period of time to learn if their virtual group proved successful, i.e., it lowers their risk of attaining a suboptimal score. This is because the time between forming a virtual group or participating in one and receiving a virtual group score can be delayed for as long as three years.
With the program primed, CMS would continue to exploit and evolve the algorithm. This means the agency would conduct on-going analysis to identify other or new solo and small group practices to join existing virtual groups and to identify year-over-year solo and small group practices that should align with other virtual groups based on their MIPS performance strengths and weaknesses. For example, virtual group A, B, and C could be advantaged in the subsequent performance year by adding solo or small group practice D. Similarly, solo or small group participant A could be further advantaged in the subsequent performance year by joining virtual group X, Y, and Z. Because MIPS participants, component measures and MIPS scoring will change year-over-year continuing to work the algorithm is in the best interest of CMS and both virtual group participants and aspirants.
CMS has stated over and again the agency’s goal is to reduce MIPS reporting burden. This motivation largely explains why CMS is proposing to again exempt approximately two-thirds of ECs from MIPS participation in 2018. However, a more aggressive virtual group approach would likely yield numerous benefits. Among others, an effective virtual group program would allow the agency to reduce the low volume exclusion thresholds. This would allow for a far greater number of solo and small group providers to participate in MIPS. Greater participation would make the MIPS program, intended moreover to improve care quality and reduce spending growth, more effective. Higher participation also means more opportunity for greater financial reward. Since MIPS is zero sum or budget neutral, more participation also means more competition among MIPS participants to improve care. More solo and small group practices with more MIPS experience and greater financial reward also means more ECs will be able to migrate to the MACRA Alternative Payment Model (APM) pathway – the ultimate goal of MACRA legislation. We strongly encourage CMS in the final 2018 MACRA rule or via other mechanisms to partner with MIPS stakeholders to develop a virtual group matchmaking model.